KR100893129B1 - System for extracting recommended keyword of multimedia contents and method thereof - Google Patents

Abstract

A system of extracting a recommendation keyword for multimedia contents and a method thereof are provided to recommend various keywords having high relationship between contents and the interest of a user according to each content reproducing time. A unit(110) for generating recommended keywords for respective contents comprises a unit(111) for extracting persons who are expected to see and hear contents, a recommendation keyword generating unit(112) and contents-keywords time synchronization unit(113). A DB server(120) comprises a seed key word DB(121), a DB(122) for persons who are expected to see and hear contents and a DB(123) for recommended keywords for each content. In case a broadcasting program is received through a broadcasting receiver(10). A keyword retrieval for the broadcasting program is concentratedly generated by a user equipment(11). A search engine receives a search query. A search query log DB stores the search keyword, time and user Identification number. The recommendation keyword generating unit collects search words of the persons who are expected to see and hear contents by referring to the DB(122) for the viewing-expected persons with respect to search queries which will be generated later.

Description

System for extracting recommended keywords of multimedia contents and method thereof {SYSTEM FOR EXTRACTING RECOMMENDED KEYWORD OF MULTIMEDIA CONTENTS AND METHOD THEREOF}

The present invention relates to a system and method for extracting recommendation keywords related to specific multimedia contents and forming a DB. More specifically, the present invention relates to a system and a method that can utilize a realistically reliable keyword collected from viewers who anticipate content broadcasted in a specific time period to be used as a recommendation keyword.

Recently, as the boundaries between broadcasting and data communication have been broken down, interactive broadcasting services such as video on demand (VOD), data broadcasting, T-commerce, and IP TV have been actively developed.

Such interactive broadcasting goes beyond simply providing the same broadcast contents to all users, reflects the needs of viewers by using a return channel, and enables a service to provide the contents directly selected by the viewers.

On the other hand, as the various services of the Internet are converged in broadcasting, the search service, one of the most important services of the Internet service, is also available in new broadcasting devices. For example, IP TV provides broadcast contents using an Internet IP-based network while providing a variety of existing IP-based services with broadcasting.

In particular, there is a great demand for content-related keyword searches that can immediately solve viewers' questions about broadcast content.

1 is a diagram schematically illustrating a system for retrieving supplemental information related to content of the related art.

Korean Laid-Open Patent Publication No. 2003-96362 discloses a configuration in which a radio listener or a television viewer provides supplementary information about a broadcast program received in real time.

The prior art system shown in FIG. 1 includes a broadcast receiving device 10, a user device 11, a broadcasting station 21, a channel information server 22, and a web server 23.

The broadcast reception device 10 receives broadcast content received from the broadcast station 21.

The channel information server 22 provides URL information on supplementary information for each broadcast content broadcast from the broadcasting station 21. When the viewer wants to search for supplemental information about the content that is currently being broadcast, the user device 11 requests the channel information server 22 for URL information about the supplemental information.

The channel information server 22 transmits URL information for each topic of a channel and a broadcast program to the user device 11, and the user device 11 retrieves supplemental information from the web server 23 using the URL for each topic.

Meanwhile, Korean Patent Publication No. 2002-81767 discloses a technology for providing celebrity and product information based on an electronic program guide (EPG) using a DB that stores celebrity and product information about a program currently being broadcast. It is starting.

In addition, Korean Patent Laid-Open Publication No. 2004-101235 discloses a technology for displaying a watched program together with the information by using a website including information about a program broadcasted on a television.

However, such a prior art presupposes that only information related to a program is stored in advance, but it does not provide any means for collecting a search word or a subject for such related information.

After all, if supplementary information or related information is manually collected and provided to broadcast program contents according to a manager's selection, it is practically impossible to provide highly relevant information to numerous broadcast contents. In addition, there may be various search keywords for specific content, but having such various search keywords is also limited. In addition, the related information selected by the manager is very unlikely to match the information that the viewer wants to search, and the amount of information searched by the viewer is also very limited because the information data prepared in advance is limited.

As such, a technology of providing keywords or information about content prepared in advance in synchronization with a broadcast program does not provide a high degree of correlation with the content, and there is a problem inconsistent with the viewer's interest.

The above-mentioned prior arts do not provide a solution for providing a keyword related to content that reflects correlation with content and viewers' interests, and thus retains the aforementioned problems.

In order to solve the above-described problem, some embodiments of the present invention provide a system and method for providing a keyword related to content that reflects a viewer's interest.

In addition, some embodiments provide a system and method that can recommend various numbers of keywords that are highly correlated with content.

In addition, some embodiments of the present invention provide a system method for providing a recommendation keyword having a high correlation between a user's interest and content for each playing time of the content.

In addition, some embodiments of the present disclosure provide a system and method for databaseing recommendation keywords for each content that can be compatible with various content applications.

According to an aspect of the present invention, there is provided a system for extracting a recommendation keyword for each content according to the first aspect of the present invention, including: a seed keyword DB for storing a broadcast time of a specific broadcast content and a predetermined seed keyword associated with the content; A viewer predictor extractor configured to extract a viewer predictor of the content from a user who searches for a search word from a search engine using the broadcast time and the seed keyword; A viewing predictor DB for storing the user identification number of the viewing predictor for each content; And a keyword recommendation generation unit configured to generate a keyword set by collecting keywords satisfying a condition in which a correlation degree with a content is greater than or equal to a threshold value among search terms searched by the viewer, based on the user identification number of the viewer and the broadcast time. Include.

The recommendation keyword extracting system may further include a content-keyword time synchronizing unit for matching the recommendation keyword for each playback time zone of the content on the basis of the number of searches per broadcast time zone of the keyword of the keyword set.

Here, the viewer predictor extractor may extract the user as a viewer predictor when a user who searches for a seed keyword is searched for a predetermined number of times or more at a predetermined time period including before and after a broadcast time.

Here, the recommendation keyword generator may generate the keyword set by using a ratio of the number of searches of a specific keyword searched by the viewers and the number of searches of all keywords searched by the viewers.

In addition, according to another aspect of the present invention, a method for extracting a recommendation keyword for each content includes: a) extracting a viewer group of viewers of the content using a broadcast time and a seed keyword of a specific broadcast content; b) constructing a keyword set by selecting a search word whose correlation with the content is greater than or equal to a threshold value among search words searched by the viewer for a predetermined time including a broadcast time of the content; c) extracting time-phased keywords from the keyword set using the number of time-phased searches for each keyword; And d) storing the recommended keyword for each time zone in synchronization with the playing time of the contents.

Here, in step b), the keyword set may be generated by using the ratio of the number of viewer search searches and the ratio of number of non-viewers searching for a specific keyword.

The step b) may further generate the keyword set by further using the number of viewers who have searched for the specific keyword, or further use the ratio of the number of times of searching for the keyword in a predetermined time zone before and after the broadcast time or broadcast time. To generate the keyword set.

Also, a recording medium in which a data structure for providing a recommendation keyword for each content according to another aspect of the present invention is recorded includes content identification data for identifying specific content;

Identification data for each reproduction time slot which divides the reproduction time of the content into a predetermined value; And at least one keyword data for each reproduction time slot corresponding to the identification data for each reproduction time slot, wherein the keyword data for each reproduction time slot is provided in synchronization with a reproduction time of the specific content.

By the above-described problem solving means, it is possible to provide a recommendation keyword whose viewer's interest is reflected and highly correlated with the content.

In addition, the present invention provides a recommendation keyword having a high correlation between the user's interest and the content for each playing time of the content, database the recommendation keyword for each content or provide in the form of a file so as to be compatible with various application services It is possible to do

DETAILED DESCRIPTION Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the present invention. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and like reference numerals designate like parts throughout the specification.

Throughout the specification, when a part is "connected" to another part, this includes not only "directly connected" but also "electrically connected" with another element in between. . In addition, when a part is said to "include" a certain component, which means that it may further include other components, except to exclude other components unless otherwise stated.

The term "search term" in this specification is used to mean a word or a combination of words searched by a user through a search engine.

In addition, the term "keyword" in this specification is used to mean a word or a combination of words that are likely to be associated with a particular content.

Thus, the "keyword" can be selected from the "search terms" already searched.

2 is a block diagram illustrating a recommendation keyword extraction system according to an embodiment of the present invention.

The recommendation keyword extraction system 100 largely includes a recommendation keyword generation unit 110 and a DB server 120 for each content.

The recommendation keyword generator 110 for each content includes a viewer predictor extractor 111, a recommendation keyword generator 112, and a content-keyword time synchronizer 113. The DB server 120 includes a seed keyword DB 121, a viewer predictor DB 122, and a recommendation keyword DB 123 for each content.

The recommended keyword extraction system 100 may perform an operation in cooperation with a normal search engine 200. The recommendation keyword extraction system 100 according to an embodiment of the present invention may utilize information of the search query log DB 210 that stores a log of a search query transmitted to a search engine.

Hereinafter, the operation and the connection relationship of each component will be described in detail.

When a broadcast program is received through the broadcast reception device 10, keyword search for the broadcast program is intensively generated by the user device 11 before and after the broadcast time.

Such a search query is received by the search engine 200, and in a typical search service system, the search keyword, time, and user identification number are stored in the search query log DB 210.

Here, the user identification number may be, for example, Bcookie. The Bcookie is a kind of identification number given to the user, and is a serial number given based on an IP address and a browser when the user visits the search page regardless of membership registration or login. If the Bcookie is not directly deleted by the user, since the existing Bcookie is used even when the user visits again, the Bcookie can be treated as a basic unit for identifying the user as a serial number corresponding to one PC.

Hereinafter, an embodiment of the present invention will be described by referring to a user identification number as a value identifying a user such as Bcookie. However, it will be readily understood by one skilled in the art that the user identification number can be any value that can identify the user.

The viewer predictor extractor 111 operates in conjunction with the search engine 200 and the seed keyword DB 121.

In the seed keyword DB 121, basic keywords related to a specific broadcast program, that is, seed keywords are stored in correspondence with a content name and a broadcast time. For example, the program name (time, subtitle), broadcast date and time, and the keyword of the content may be stored in correspondence. Here, the seed keyword may be a name or a variant of the broadcast content. Meanwhile, the seed keyword may be stored by dividing the keyword into categories and giving different weights. For example, in the drama category, the first category may include a title and a cast name, and the second category may include an actor name and a crew. Here, the first category having a higher search probability may have a higher weight than the second category.

The viewer predictor extracting unit 111 checks whether there is a seed keyword stored in the seed keyword DB among the search terms received by the search engine 200 within a predetermined time period before and after the broadcast time, and searches for the seed keyword. Extract the identification number. The extracted user identification number may be classified as a viewing predictor. Here, in order to extract the viewer, it may be checked whether a normal connection has occurred, and the identification number of the user who has searched for the seed keyword having a threshold number of times or more may be extracted as the viewer. In this case, when the search engine provides autocomplete in the search box, the search term based on the autocomplete is included, but the search engine that is already completed and provided on the screen by the search engine (for example, a popular search query, a suggestion query, and a real-time rising search query) ) Can be excluded from viewing viewers.

When the viewing predictor is extracted, the user identification number of the viewing predictor is stored in the viewing predictor DB 122. In the viewing predictor DB 122, user identification numbers corresponding to a specific broadcast content or a series or group of the specific broadcast content are matched and stored.

The recommended keyword generator 112 may collect a search term searched by the viewer predictor in the broadcast time zone of the corresponding content by referring to the viewer predictor DB 122 for a search query generated later. In addition, the recommended keyword generator 112 determines the keyword as a collected keyword when the collected keyword satisfies a specific condition and includes the corresponding keyword in the keyword set. The specific condition may be a condition that a correlation between the corresponding content among keywords is equal to or greater than a threshold.

For example, the ratio of the number of searches during the broadcast time of the viewer for the keyword may be considered in the specific condition. In addition, a condition in which the number of users searched for the keyword should be greater than or equal to a predetermined value or the number of searches for the keyword should be concentrated in a specific time zone (for example, a broadcast time of the corresponding content).

The recommended keyword generator 112 determines a keyword satisfying the above-described condition as a recommended keyword for the corresponding content, and generates a keyword set for the corresponding content.

The content-keyword time synchronizing unit 113 performs a function of matching keywords included in the keyword set for each reproduction time zone of the content. For example, at 60-second intervals, a keyword having a large ratio for each time zone is selected based on the number of times of keyword search for each time zone. Here, the predetermined inappropriate keyword is filtered out. For example, remove inappropriate keywords as keywords such as live, rebroadcast, show, organize, airtime, etc. The search query log DB 210 also includes search time information for keywords selected as recommended keywords. Therefore, when the broadcast start time and the search time of the content are combined, the content playback time zone and the search time can be synchronized.

Finally, the keywords matched by content are stored in the recommended keyword DB 123 for each content. The data stored in the recommended keyword DB 123 for each content are matched with the recommended keyword synchronized with the playing time of each content. Therefore, when specific content is provided through various services, data stored in the recommended keyword DB 123 for each content may be provided together as a separate file, and the viewer may view the content and play the content using the file. And recommended keywords synchronized with.

3 is a diagram illustrating a data structure stored in the seed keyword DB 121 according to an embodiment of the present invention.

The seed keyword DB 121 stores a broadcast time and a seed keyword corresponding to each content. For example, as the broadcasting time and seed keyword of one X-file, the most basic keywords such as "X-file" and "alien" are stored.

As described above, since a keyword related to a specific content is rapidly searched for before and after a broadcast time, a user who searches for a seed keyword at a predetermined time zone before and after the time when the first episode of the X-file is aired may read the first episode of the X-file. It is judged as a viewing predictor who is watching.

When the seed keyword DB 121 is constructed for each broadcasted content, candidates for viewing predictors for each content may be collected according to an embodiment of the present invention.

4 is a diagram illustrating a data structure stored in the audience predictor DB 122 according to an embodiment of the present invention.

Through the above-described process, when the viewing predictor is extracted, the user identification numbers UID 1, UID 2, and UID 3 of each viewing predictor are matched with the content and stored.

When the viewer is extracted, the search term searched by the viewer in the broadcast time zone of the same type of content in the future will be a candidate of the recommended keyword corresponding to the content.

5 is a diagram illustrating a data structure of a recommendation keyword DB 123 for each content and a process of generating the same according to an embodiment of the present invention.

Since the user identification number for each content stored in the viewing predictor DB 122 can identify the viewing predictor of the corresponding content, the recommended keyword is analyzed through the analysis of the search query log stored in the search query log DB 210. can do.

Since the search query log DB 210 stores a log about a user identification number, a search time, and a search keyword related to the search query generated by the search engine, the search predictor may extract the keyword and time searched by the viewer.

For example, a viewer (UID 1) performed a keyword search for “virus”, “FBI”, and “Scully” at the “X-File 2” broadcast time, and the keyword satisfies a specific condition to determine the recommended keyword. If included in the set, it may be stored in the recommended keyword DB 123 for each content.

Here, "X-File 2" began to air at 20:00, "virus" searched at 20:30, "FBI" searched at 20:40, and "Scully" searched at 20:45 Respective keywords can be adopted as one of the recommended keywords of 30 to 31 minutes, 40 to 41 minutes, and 45 to 46 minutes of the playback time, respectively.

As described above, when the recommendation keyword DB 123 for each content stores the recommendation keyword for each content for each content, a communication-broadcasting convergence service such as IP TV, web TV, etc., reflects user's interests for each content. It is possible to provide a recommended keyword. The keyword may be used in an environment in which text input is inconvenient (for example, by using a remote controller), so that a user may easily select a desired search word or may be used in an autocomplete function operated when a part of the search word is input.

Each component of the above-mentioned recommendation keyword extraction system may be implemented separately, but those skilled in the art will readily understand that the components of the recommended keyword extraction system may be integrated and implemented on one piece of hardware. In addition, it will be easily understood that one embodiment of the present invention is implemented as an integrated part of the search engine 200 having the recommended keyword extraction function without departing from the spirit of the present invention.

6 is a flowchart illustrating a recommendation keyword extraction method according to an embodiment of the present invention.

In extracting the recommended keyword according to an embodiment of the present invention, first, a viewer predictor group for a specific broadcast content is extracted and the viewer predictor group is stored in correspondence with the content (S100).

When the viewing predictor is extracted, a keyword set is constructed from the search word searched by the viewing predictor (S200). The keyword set may include a search word that satisfies a specific condition determined to be highly related to the corresponding content among search words searched by the viewer for the broadcast time of the same broadcast content or a series of content.

Thereafter, when the keyword set is configured, the keyword for each content reproduction time zone is extracted (S300). Using the search time of the search word constituting the keyword set, it is possible to determine in which section of the broadcast content reproduction time the keyword corresponding to each search word is performed. Therefore, the recommended keywords constituting the keyword set may be matched with the playing time section of the content.

In step S400, the time-synchronized recommendation keyword for each content is made into a DB. The DB-recommended keyword may be used as a separate file or provided in synchronization with content through a specific server.

Hereinafter, the steps S100, S200, S300, and S400 illustrated in FIG. 6 will be described in more detail.

FIG. 7 is a flowchart illustrating a method of extracting and storing a viewing audience group in detail according to an embodiment of the present invention.

In order to extract the viewing predictor, a seed keyword corresponding to a broadcast time and content is first constructed (S110). The seed keyword may be a group of the most basic keywords representing content. For example, the content may be a name, performer, or theme. The broadcast time is information about a time when the content is broadcast, and may also include information about a date. In addition, a predetermined time zone including a predetermined time (for example, 30 minutes) before and after the broadcast time may be included in the broadcast time.

In step S120, it is determined whether a search generated at the broadcast time or a predetermined time period is a normal search. The abnormal search such as the browser being closed before the search result is transmitted ends the process (S170) and is not considered in the extraction of the viewer predictor. The determination of whether the search is normal may be made by referring to a variable value that determines whether the browser is normally connected to the search result page. In addition, even if a search using a search word already completed on a page provided by a search engine is not considered for extracting viewers.

In step S130, it is determined whether the seed keyword in a predetermined time zone before and after the broadcast time has been searched. When a specific user searches for seed keywords, proceed to the next step.

When the seed keyword is searched, it is determined whether the user who searched the search keyword has performed the search above the threshold (n), and when the search is above the threshold (n), the user proceeds to the next step (S140). ) ..

Finally, it is determined whether the user to be registered as the viewing predictor can be identified as the user identification number (S150). For example, if the value of Bcookie is “-”, there is no Bcookie and it is impossible to identify the user.

Although not shown in FIG. 7, when one user inputs a search word too many times or frequently during a broadcast time, it is preferable to exclude the user from the viewer.

In addition, as described above, when the weight of the category of the seed keyword is assigned, it is also possible to classify the seed keyword as a searched user when the sum of the category weights of the searched seed keywords becomes equal to or greater than a predetermined value. According to the above embodiment, it is possible to exclude a user who performed an irrelevant search from a viewer by assigning a low keyword to a seed keyword related to a duplicate appearance and a specific issue. For example, when an actor who is appearing in a large number of contents or an actor who is interested in a social issue accidentally is a performer of the content, the seed keyword for the actor name is given a low weight so that the current broadcast content Do not extract a user who does not watch as a viewing predictor.

If the above determination process is satisfied, the user who performs a search for a keyword related to the content while watching a broadcast for the corresponding content is identified as the viewing predictor, and the identification number of the identified user is registered in the viewing predictor group (S160). . The user identification number of the viewing predictor group is later used to identify viewers of the corresponding content or series of the same content.

8 is a flowchart specifically illustrating a method of constructing a keyword set in an embodiment of the present invention.

When the viewer predictor group corresponding to the content is established, the recommended keyword for each content is collected using the search results of the viewers.

In step S210, the search result of the viewing predictor for a predetermined time is collected. For example, the predetermined time may be 30 minutes before and after a broadcast time of specific content. Here, for each search term searched by the viewer, it is determined whether the keyword can be a recommended keyword of the corresponding content.

In step S220, it is determined whether the viewer search rate ratio is greater than the threshold value K for the keyword. Here, the determination of the viewer search frequency ratio may be performed through Equation 1 as follows.

(QC A / QC A-total) / (QC B / QC B-total)> K (where K> 0)

Here, the meaning of each parameter is as follows.

QC A: Number of times viewers searched for this keyword

QC A-total: viewership search queries

QC B: Number of searches for that keyword by non-viewers

QC B-total: Search queries for all non-watching predictors

That is, when the ratio of the number of searches generated between the viewers who are viewing viewers is higher than the ratio of the number of searches generated by the non-viewing predictors for a specific keyword, the corresponding keyword has a high correlation with the content.

In step S230, it is determined whether the number of viewers who have searched for the keyword is larger than the threshold m. That is, the keyword proceeds to the next step only when a search is performed by at least m or more viewers.

On the other hand, it is possible to calculate whether the viewers who searched for a specific search word are extracted through a predetermined number of seed keywords or more, and to include the search word in the keyword set only when it is extracted through the predetermined number or more seed keywords. This is because a search term searched by viewers extracted through a single seed keyword may be a case of a search term having a different meaning including spelling of the seed keyword.

For example, if one of the seed keywords is “file,” the viewer who searched for “profile” may be extracted as the viewer predictor through the seed keyword, and thus the viewer predicted through the single seed keyword “file”. It is desirable to exclude only the search terms from the keyword set.

In step S240, it is determined whether the ratio of the number of searches for the predetermined time of the corresponding keyword is larger than the threshold value P (where 0 < P < 1). That is, when the ratio of the number of times of searching for the corresponding keyword among the broadcasting times to the number of searches for the whole day is larger than the threshold P, it may be determined that the corresponding keywords are concentrated in the broadcasting time.

If the above condition is satisfied, the keyword is stored in the keyword set as a recommended keyword for the content (S250).

The above-described steps S220, S230, and S240 may be sufficiently performed using the information stored in the seed keyword DB 120, the viewer predictor DB 122, and the search query log DB 210 described above. There will be.

In addition, even if the formula or condition specific to the step (S220), step (S230), step (S240) is replaced with other conditions that determine the correlation with the content using the number of searches, the number of searchers, the search time, Those skilled in the art will readily understand that the invention does not depart from the overall gist.

9 is a flowchart specifically showing a method of extracting a recommendation keyword for each content playback time zone according to an embodiment of the present invention.

As described above, if a recommended keyword set for a specific content is configured, the keyword itself is useful enough, but user convenience will be further maximized if it is synchronized with the theme and interests of the content playing time.

In step S310, the number of searches for each time zone is calculated. That is, the number of times a keyword is searched is divided by a specific time interval (for example, 60 seconds) to obtain a number of searches for the keyword.

The calculated number of searches for the corresponding keyword is calculated by calculating a ratio with respect to the number of searches in the broadcast time (S320). Here, instead of the ratio of the number of searches in the broadcast time, the ratio of the number of searches in the broadcast time + A may be used (A> 0). If it is not greater than the threshold Q, it is excluded from the keyword of the corresponding time zone (S360). If it is greater than the threshold Q, the process proceeds to the next step.

Here, without using the threshold Q, it is also possible to select a predetermined number of upper keywords having a high number of search times as recommended keywords for each time slot.

In step S330, it is determined whether the corresponding keyword is an inappropriate keyword. Here, the inappropriate keyword is a predetermined keyword irrelevant to the content and may be a keyword such as "live broadcast", "re-broadcast", "review", and the like.

In step S340, the search frequency ratio of each time zone is high, and keywords that are not inappropriate keywords are registered as the corresponding time zone keywords.

When the recommended keyword is registered in every time zone, the broadcast time is synchronized with the content playback time (S350). The recommended keyword selected for each time zone from the start time of the broadcast time matches the recommended keyword for each time zone from the playback start time.

10 is a diagram illustrating a scene in which a recommendation keyword is utilized according to an embodiment of the present invention.

FIG. 10 illustrates a screen when 10 minutes have passed since the start of playback of a specific content (eg, an X-file). In the screen, a person corresponding to the main character, a car in which the main character is boarded, and a UFO are displayed.

Recommended keywords corresponding to the screen of the content are, for example, "X-file", "X-file hero", "X-file ending", "Alien", "UFO", "David Duchovny (real name of the X-file hero)" , “Ex-Pile Car”, “Ferrari”, and the like. These recommended keywords may be provided together with the screen or separately at the request of the user.

The keyword may call a search result through a search engine in a service provided with Internet search such as IP TV. In addition, even in an input tool that is inconvenient to input characters such as a remote controller, a user can easily search by selecting a desired keyword.

In addition to the above-mentioned applications, the above recommended keywords may be used for searching in a web TV environment in which Internet communication is readily available, and various multi-media services regardless of the medium may be used when the recommended keyword file is synchronized with the multi-media file of contents. Could be applied to

In addition, by providing specific information or data about the keyword to the recommended keyword can be utilized for business purposes, on the contrary, by providing a result of searching a simple keyword through a plurality of search engines to maximize the convenience of users who require a variety of information It may be.

As described above, the recommendation keyword extraction system and method according to the embodiment of the present invention may have various correlations with content and may be variously applied to any service in that it may reflect the interests of viewers.

An embodiment of the present invention may also be implemented in the form of a recording medium including instructions executable by a computer, such as a program module executed by the computer.

In addition, a data structure according to an embodiment of the present invention may also be stored in a recording medium executable by a computer.

Computer readable media can be any available media that can be accessed by a computer and includes both volatile and nonvolatile media, removable and non-removable media. In addition, computer readable media may include both computer storage media and communication media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Communication media typically includes computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave, or other transmission mechanism, and includes any information delivery media.

Although the methods and systems of the present invention have been described in connection with specific embodiments, some or all of their components or operations may be implemented using a computer system having a general purpose hardware architecture.

The foregoing description of the present invention is intended for illustration, and it will be understood by those skilled in the art that the present invention may be easily modified in other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, it should be understood that the embodiments described above are exemplary in all respects and not restrictive. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as distributed may be implemented in a combined form.

The scope of the present invention is shown by the following claims rather than the above description, and all changes or modifications derived from the meaning and scope of the claims and their equivalents should be construed as being included in the scope of the present invention. do.

1 is a diagram schematically illustrating a system for retrieving supplemental information related to content of the related art.

2 is a block diagram illustrating a recommendation keyword extraction system according to an embodiment of the present invention.

3 is a diagram illustrating a data structure stored in the seed keyword DB 121 according to an embodiment of the present invention.

4 is a diagram illustrating a data structure stored in the audience predictor DB 122 according to an embodiment of the present invention.

5 is a diagram illustrating a data structure of a recommendation keyword DB 123 for each content and a process of generating the same according to an embodiment of the present invention.

6 is a flowchart illustrating a recommendation keyword extraction method according to an embodiment of the present invention.

FIG. 7 is a flowchart illustrating a method of extracting and storing a viewing audience group in detail according to an embodiment of the present invention.

8 is a flowchart specifically illustrating a method of constructing a keyword set in an embodiment of the present invention.

9 is a flowchart specifically showing a method of extracting a recommendation keyword for each content playback time zone according to an embodiment of the present invention.

10 is a diagram illustrating a scene in which a recommendation keyword is utilized according to an embodiment of the present invention.

Claims (22)

A seed keyword DB for storing a broadcast time of specific broadcast content and a predetermined seed keyword associated with the content; A viewer predictor extractor configured to extract a viewer predictor of the content from a user who searches for a search word from a search engine using the broadcast time and the seed keyword; And A recommended keyword generation unit for generating a keyword set based on a user identification number of the viewer and the broadcasting time, and collecting a keyword satisfying a condition that a correlation between the content is greater than or equal to a threshold among search terms searched by the viewer; Suggested keyword extraction system that includes. The method of claim 1, And a content-keyword time synchronizing unit for matching the recommendation keyword of each of the playback time zones of the content based on the number of times of search for each time zone of the keyword of the keyword set. The method of claim 2, And the content-keyword time synchronization unit removes a predetermined inappropriate keyword among the keywords included in the keyword set. The method according to claim 2 or 3, The content-keyword time synchronizing unit, the recommended keyword extraction system for storing the data related to the keyword recommended for each playback time zone to be synchronized with the content. The method of claim 1, And a viewing predictor DB for storing the user identification number of the viewing predictor for each content. The method of claim 5, And the viewing predictor extracting unit extracts the user as a viewing predictor when a user who searches for a seed keyword in a predetermined time zone including a broadcast time searches for a predetermined number of times or more. The method of claim 5, And the recommendation keyword generation unit generates the keyword set by using a ratio of the number of searches of a specific keyword searched by the viewers and the number of searches of all keywords searched by the viewers. The method of claim 7, wherein And the keyword recommendation generation unit generates the keyword set by further using the number of viewers who have searched for the specific keyword. The method of claim 7, wherein And the recommendation keyword generation unit generates the keyword set by further using a ratio of the number of times of searching the keyword during the broadcasting time to the number of times of keyword search for the whole day. a) extracting a viewing audience group of the content by using a broadcast time of a specific broadcast content and a seed keyword previously stored in association with the content; And b) constructing a keyword set by selecting a search word whose correlation with the content is greater than or equal to a threshold value among search terms searched by the viewer for a predetermined time including a broadcast time of the content; Suggested keyword extraction method comprising a. The method of claim 10, c) extracting time-phased keywords from the keyword set using the number of time-phased searches for each keyword; And d) storing the recommended keyword for each time zone by synchronizing with the playing time of the contents; Suggested keyword extraction method further comprising. The method of claim 10, In the step a), during the predetermined time including the broadcast time of the specific content, extracting a user who searched for a search word including the seed keyword more than a predetermined number of times as a viewing predictor. The method of claim 12, The seed keyword corresponds to a category having a different weight, and when the sum of the weights of the categories of the seed keywords included in the search terms is equal to or greater than a predetermined value, the user searching for the search term is recommended as a viewer. How to extract keywords. The method of claim 10, In the step b), the keyword set is generated using the ratio of the number of viewer predicted searches and the ratio of non-viewed audience searches for a specific keyword. The method of claim 14, In step b), the keyword set extraction method further generates the keyword set by further using the number of viewers who have searched for the specific keyword. The method of claim 15, And a viewer predictor searching for the specific keyword is a viewer predictor extracted through a predetermined number or more of seed keywords. The method of claim 14, The step b) is to generate the keyword set further using the ratio of the number of times of search for the keyword in a predetermined time zone including the broadcast time. The method of claim 11, In the step c), by dividing a broadcast time into a predetermined time interval to extract the recommended keyword for each time zone by using the value or rank of the number of times of search for a specific keyword for each time interval. A computer-readable recording medium having recorded thereon a program for causing a computer to execute the steps according to any one of claims 10 to 18. A recording medium having a data structure recorded thereon, Content identification data identifying specific content; Identification data for each reproduction time slot which divides the reproduction time of the content into a predetermined value; And At least one keyword data for each playback time zone corresponding to the identification data for each playback time zone; Wherein the keyword data for each reproduction time slot is provided in synchronization with a reproduction time of the specific content. The method of claim 20, The keyword data is generated from a search word that satisfies a condition that the correlation with the content is greater than or equal to a threshold value among search terms of a viewer predicted among users who search for a search word including a seed keyword during a broadcast time of the content. Recording medium in which a data structure is recorded. The method of claim 21, And using the keyword search time of the keyword data and the broadcast time of the content to correspond the keyword data for each playback time zone and the identification data for each playback time zone.

Images